PYB350 Advanced Statistical Analysis PDF

Preview

Summary

Hey guys, I hope this helps you all out!
These are notes on all the lectures & tutorials....

Description

PYB350: Advanced Statistical Analysis Workshop 1

Lecture 10 - Is all the content you learn.

First Step to Conduct Research Identify phenomena of interest & why important to research this phenomena. E.g. Student wellbeing, student stress & anxiety increasing why? Identify broad factors relevant to this topic (e.g. workload, financial stress, social support, online learning). Read the scientific literature until you understand the current state of knowledge in these areas. If there’s an established theory relevant to the topic, ask what this theory predicts about the phenomena. If there’s no clear theoretical perspective, ask what evidence is needed to allow a theory to be developed. If there’s competing theoretical perspectives, ask what evidence is needed to establish which theory is correct. Formulate the Research Question. Identify best research method to research penehomena Main Objectives of Psychological Research Description: Portraying the phenomenon accurately. E.g. Piaget’s Theory of child development arose from detailed observations of his own children. Explanation: Identifying the cause(s) of the phenomenon. E.g. The effect of cognitive training on problem-solving ability. Prediction: Identifying risk factors of a phenomenon can help predict when it might happen. E.g. What factors predict/are associated with student stress. Your methodological approach depends on the nature of your research question. Major Epistemological Approaches Positivist or Etic: Concerned with uncovering generalisable patterns & laws based on objective empirical data (tends to be deductive in nature). Quantitative methods are predominately positivist. Interpretivist or Emic: Concerned with subjective interpretation, context, specific, not concerned with generalisability but with deep understanding in line with inductive approaches. Qualitative methods are predominately interpretivist. Major Research Approaches Quantitative Studies: Collect numerical data from a large number of participants to draw generalisable conclusions e.g. IQ scores on psychological tests, ratings on attitude scales. Qualitative Studies: Collect non-numerical data from a small number of participants to answer in-depth research questions. E.g. Interview data, pictures, documents, observations. Multi Methods: Use a combination of qualitative & quantitative research methods. E.g. Quantitative relationship found & qualitative enquiry used to understand the process underlying the relationship. Main Quantitative Methods Survey Designs: Often cross-sectional & primarily descriptive. E.g. 50% thought that...Typical analyses descriptives mean, frequencies, cross tabs. Correlational Designs: Primarily for prediction but causality cannot be assumed, typical analysis correlation & regression. Quasi Experimental Designs: Where IV’s naturally occurring (e.g. age group) show relationships between IV & DV’s, typical analysis ANOVA. Experimental Designs: Where IV is manipulated & extraneous variables controlled, typical analysis ANOVA, ANCOVA. Longitudinal Designs: Measures change over time. E.g. The NZ Attitudes & Values Study (NZAVS) is a 20 year longitudinal national study of social attitudes, personality & health outcomes. Qualitative Research An interpretive research approach that relies on subjective data i.e. participant’s subjective perspectives in their natural environment. Concerned with understanding world views & lived experience rather than testing overarching generalised laws of human behaviour. Main Qualitative Methods: Observations, Interviews, Focus Groups, Text document/analysis.

Ethical Research National Health and Medical Research Council has a National Statement on Ethical Conduct in Human Research. Principles: Research Merit & Integrity; Justice; Benefiance & Respect. Every institution that conducts research with human participants is required to have a Human Research Ethics Committee (HREC’s) which examined proposed research projects & ensure they meet the National State Guidelines. These guidelines are designed to protect participants & ensure the social benefits of the research outweighs any risks to participants. Ethical dimension to the relationship between science & society e.g. Funding of research by the government/corporations (e.g. drug company) must be independent or they may be a conflict of interest. HREC’s will request a conflict of interest statement. APS Standard 8 - Research & Publication All research must have approval by Institutional Ethics Committee. Key elements of ethical research: Participation only with informed consent.

Fully informing participants of all aspects of the study, purpose & procedures, risks & benefits. Participants can make an informed decision to participate or decline. Freedom to withdraw (at any time with no comment). Deception only with full debriefing. Confidentiality or anonymity of data provided by participants. 10 Steps in Conducting Research 1.Identify phenomena of interest & why research is needed. 2.Identify what broad factors are relevant to understanding the topic. 3.Gather existing information & evidence (e.g. past research & theory). 4.Statement of research question & if deductive formulate research hypothesis. 6.Identify research methods appropriate to the nature of your research question. 7.Ethical consideration & approval. 8.Data collection & analysis. 9.Interpretations & conclusions. 10.Reporting the study. The value, integrity & validity of the research depends on the rigor applied to each of these steps. This unit focuses on steps 8-10. Note: Appropriate reporting & interpretation of data is an aspect of ethical research. Paradigms in Social Research Research spectrum including realism & constructionism.

Designing Qualitative Research

Two Predominant Frameworks Interpretive Framework: Trying to understand & interpret some phenomena of interest. Might try to understand what the circumstances in which nurses are presumed to be women. Critical Framework: Critique a version of society. May bring in concepts of the patriarchy to understand why there are more female nurses.

Obtrusive may include more individual involvement of a subject/s e.g. Interviews, Focus groups.

Analytic Approaches

Thematic Approaches

Thematic Analysis Braun & Clarke (2006) Using Thematic Analysis in Psychology, Qualitative Research in Psychology 77-101. Foundational method for qualitative analysis. Thematic analysis techniques are used (or are transferable for use) in a range of thematic approaches. Generic & flexible: Can be used for research across the realist-constructivist continuum. Can be descriptive (semantic themes) or interpretive (latent themes). Method for identifying patterns (i.e. themes) within data. Theme: Some level of patterned responses within the data. Researcher judgement is necessary to determine a theme & identify its boundary. Latent Themes: Below the service. May look at a patient from a motivational level e.g. are they practising resistance? Semantic Themes: Surface level. May look at patient preferences/experiences/what they don’t like. Deductive Research: Often is a hallmark of quantitative research, but is sometimes in qualitative research (usually theory-driven in this case e.g. theory of patriarchy). Phase 1: Becoming familiarised with the data. Immersion both breadth & depth. For interviews, it is ideal to use recordings as well as transcripts. Time-consuming (hence smaller sample sizes than quantitative research). Phase 2: Generalising initial codes. Identifying features within the data: Semantic (i.e. surface meaning) & Latent (i.e. meaning that’s indicated but not directly apparent). Code as broadly as possible (i.e. try not to dismiss anything at this stage). Data can be coded more than once. If you aren’t fully immersed in your (qualitative) data the chances of producing a robust analysis are relatively slim. This is commonly why we work with smaller sample sizes in qualitative data. Phase 3: Searching for themes. Sort codes into candidate themes & sub-themes. (Theme/Superordinate theme/Higher-order themes mean the same thing). Phase 4: Reviewing Themes. Some themes might be rejected (too little supporting data or data are too diverse). Some themes might be collapsed (perhaps becoming sub-themes). Phase 5: Defining & Naming Themes. Identifying the essence of the theme. Determine what makes it different from the other themes. Understand how it relates to other themes. Activity 1: Interview with a Brain Tumour Patient People w/brain tumours often had high care needs despite this, many did not access supportive care services that could help meet their needs. It could not be found out why so patients who weren’t accessing these services were interviewed. Replicating Phase 2 of Thematic Analysis. Textbooks Encouraged to do Line by Line Coding, where you look into the text on a very detailed level - breaking the chunk of text into pieces. Rather you could code by units of possible meaning. To help this process it’s best to annotate the document.

Semantic Example of Coding

The theme below is Information. Dynamic relationship between subcategories regarding information & would like printed information.

1.Didn’t Want Supportive Care Services: Didn’t think they needed help (subcategory) in Fragments 3,4, & 6. Used own/Informal Resources in Fragment 2, 8, & 11. 2.Barriers to Supportive Care Services: Incapacity: Tumour/Treatment in Fragments 6,7, & 10. Wasn’t Aware of Help in Fragments 1, 5, & 9. Content Analysis: Straddles qualitative-quantitative divide. Data analysed qualitatively but reported quantitatively. Aims to specify a clearly defined set of content categories. These can be identified: Deductively (i.e. prior to analysis). Inductively (i.e. through analysis). Coding manual developed to reliably code data. Commonly do tests of inter-rater agreement. Grounded Theory Methods Analytic approaches designed to develop theories from data. Methods include Simultaneous data collection & analysis. Check & refine emerging analysis through the collection of new data. Purposive → Theoretical Sampling: Started developing a candidate theory of your phenomena of interest which has changed your sampling strategy. Constant Comparison: E.g. Checking to ensure that a candidate analysis/theory accounts for all relevant cases. Saturating Theoretical Concepts: New data doesn’t necessitate the revision of a candidate analysis/theory. How a researcher decides when to stop collecting more data, over a period of time/number of cases they may not learn anything new and stop collecting new data that’s repetitive.

Interpretive Phenomenological Analysis Detailed examination of the personal lived experience. Wishes, desires, feelings, motivations, beliefs, behaviour, action. Treats participants as experiential experts. Develop an understanding of a phenomenon from the perspective of the person experiencing it. Aims to go beyond what people say to understand deeper meaning. Seeks to identify the different parts of the experience. Interprets data using the hermeneutic circle. Examine the whole in relation to its parts, the parts in relation to the whole, considering the contexts in which the whole & parts are embedded. Adopts an idiographic approach. Benign with the particular (sample sizes tend to be small in IPA research). Ensure any generalisations are grounded in these particulars. Limitations of Coding-Based Analysis Coding is a concept that’s been appropriated from quantitative research. This assumes things like: Phenomena can be split up & coded into parts. Coded parts can be considered out of their surrounding context. Coded parts remain meaningful once contextualised. People who may not have been closely involved in the data generation process can meaningfully contribute to inter-rater agreement coding. Limitations of Interview-Based Research Failure to consider interviews as social interactions. Interviewer’s questions are routinely removed in study reports. Flooding the interview with a social science agenda & categories. Interviews require people to adopt a social scientific perspective on their own experience. E.g. “Do you feel that there are constraints on teachers' time & the resources that are available to you…” Complex problems of stake & interest. Interviewees are routinely treated as neutral informants on their own practices, rather than someone with a variety of stakes & interests in a matter (e.g. appearing to be a reasonable person). E.g. Social desirability/Interviewees wanting to seem likeable may influence the data given. Reading: Using Thematic Analysis in Psychology, Braun & Clarke (2006) Thematic analysis is a widely used qualitative analytic method within psychology. It is theoretically flexible to analysing qualitative data. Thematic analysis is a method for identifying, analysing & reporting patterns (themes) within data. It minimally organises & describes your data set in (rich) detail. Workshop 2- Qualitative Research Methods: Language-Based Approaches Qualitative Research Methods in Psychology Thematic Analysis Compared to Conversation Analysis - Data TA: Often interviews. CA: Always naturalistic social interaction. Analytic Focus: TA: Experiences. CA: Practices. E.g. What’s communicating with clinicians like for families when their child has a life-limiting (terminal) condition? Relatively small population. Diverse conditions, prognoses, symptoms, treatments, ages, social competencies, etc. Two ways of answering this question - focus on: Experiences of communication (Thematic Analysis). Practices of communication (Conversation Analysis). Communication Experience

Communication Practice

Introduction to Conversation Analysis Leading approach to the study of human social interaction. Conversation analysis is both: A field of research, Transdisciplinary: Sociology, Linguistics, Anthropology, Communication studies, Psychology, etc. And a unique method.

Key Theoretical Assumptions Social interaction is orderly - even in its minute details. Order at all points (Sacks). This order is a result of shared methods of reasoning & action. All (competent) members of society share access to these methods. These methods are identifiable through careful & detailed analysis of real-world social interaction. “...a naturalistic observational discipline that could deal with the details of social action(s) rigorously, empirically, & formally.” (Schegloff & Sacks, 1973). Influences: Harold Garfinkel - Ethnomethodology highlights that we can use member’s methods as analytic tools: Members are analysts too. Some people argue conversation analysis is part of the Ethnomethodology enterprise/a subfield of it. Members Methods: Methods people use in everyday social life. Erving Goffman - The Interaction Order - How we interact with others in everyday social life. Distinctive Methodological Features: “We’ll be using observations as a basis for theorising...from close looking at the world we can find things we couldn’t by imagination assert were there” (Sacks, 1984, 25). Data: Recordings of naturally-occurring social interaction (e.g. Can’t use field notes or an Interview). Data analysed in their raw form (i.e. No processing that transforms the data, like a semi-structured recorded interview that is transcribed by a transcriber). Analysis: Qualitative & Inductive. Case-by-case analysis → Patterns → Generalisations. You need generalisations that accommodate for differences between cases but also account for the similarities. “...But without allowing them to congeal into an aggregate” (Strivers & Sidnell, 2013:2).

(Japasonian Transcription)

2hr (Time Stamp) Conversation Analytic Methods Detailed inspection of single cases - collection of related practices. Noticing something distinctive in social interaction. Unmotivated examination allows you to notice distinctive things like laughter in sad situations. Why that now? E.g. Why laugh in the face of death? Build collection of related practices. Practice: Distinctive character. Specific location/environment of production (e.g. in particular type of turn/sequence). Distinctive in its consequences (e.g. the action it implements). These social practices are often beyond the level of conscious awareness. Hence avoidance of self-report methods. Sacks, Schegloff & Jefferson “Proof Criterion” (Participant’s understandings).

“Since it’s the parties’ understanding of prior turns’ talk that is relevant to their construction of next turns, it’s their understandings that are wanted for analysis” (Sacks et al., 1979: 729). Next Turn Proof Procedure - Evidence/Support you can see from a particular behaviour may be seen in the next turn. If A occurs then B can happen. Conversation analysis requires data-internal evidence. Empirically grounding analysis is the conduct of participants (Schelgloff, 1997). Identifying how a practice is procedurally consequential for an interaction (Schlegoff, 1992). Conversation Analytic Methods Deviant Case Analysis: Cases that deviate from the pattern. “Incongruous cases in which the proposed regularity wasn’t realised”. Either: The analysis needs to be refined. The analysis is correct & participants own conduct orients to the unexpected nature of this instance. The exception that proves the will (The analysis is correct & the participants own conduct is unexpected/deviant). The goal of Conversation analysis is to explain what can happen in both ordinary & extraordinary circumstances. Conversation Analysis & Quantification “CA is arguably the most quantitative of the qualitative social science methods”.(Stivers,2015: 3). Like quantitative research, seeks to make generalizations. Distributional evidence is a key part of the CA’s method. E.g. “...massively in conversation, references in reference occasions are accomplished by the use of a single reference form” (Sacks & Schegloff, 1979: 17). Small but increasing number of studies using CA to develop coding criteria. Activity: Therapy session. Observe how the client tells the story about another young girl rather than her own story. What did she do with the story? She related the story back to her own situation & family. E.g. Green represents repetition.

Example: Managing Expectations About Psychotherapy

The Second example gives a clearer understanding of what will happen in the sessions.

This one has no expectation management (below).

Coding for these examples.

‘Substantial’ Inter-rater Agreement (Weighted Kappa = 0.78; 95% CI 0.52 to 0.94). Relationship between expectation management & retention. Model adjusted for: Age, Gender, Trial centre, Beck Depression Inventory (BDI) score, use of antidepressants, counsellor at GP practice, & difficulties during initial moments of first session. Clients exposed to Type 2 expectation management remained in therapy for an average of 1.6 sessions longer than other participants (95% CI: 0.2 to 3.1). Both a denominator & numerator is necessary to quantify frequency occurrence in social interaction. Numerator = Focal Practice. “Whose presence should count as events & whose nonoccurrence should count as absences” (Schegloff, 1993: 103). Denominator = Environments of possible relevant occurrence. Consider: Managing expectations about psychotherapy vs empathy in psychotherapy. “We have as yet little compelling evidence that there’s distinctive payoffs to be derived from quantitative analysis that would counterbalance the considerable analytic pitfalls that lurk about” (Schegloss, 1993). Workshop Three (Little videos rehash this content, youtube videos uploaded to BB). Compare Means (Top tab) → One Sample T Test (E.g. To compare how well one did in a test).

Df is N - 1, so in this case N = 16 - 1 = 15 (Df) We have a less than 0.36 chance of a mean of 60 coming from the same distribution as the normal distribution of Reading Scores of M=54. Confidence intervals around the difference of our Mean of 60 to the population Mean is 6. We are 95% sure that the real difference would lie between .44 and 11.56 Null Hypothesis: Cut-off value to reject the null hypothesis is 0.05 (5%). T test is less than a 5% chan...